Data Source and Study Design
The analysis is based on the nationally representative study of the Multiple Indicator Cluster Survey (MICS-2018). The study was conducted by the Lesotho Ministry of Development Planning through the Department of the Bureau of Statistics in collaboration with the United Nations Children's Fund (UNICEF) as part of the Global MICS program. MICS covered several modules and indicators across households, children (0-17), women (15-49 and males (15-49) across all 10 districts of Lesotho and all ecological zones (highlands, lowlands, foothills and the Senqu Valley). The data was collected between April 2018 to September 2018.
Lesotho MICS is a multistage stratified sampling survey. The sample was designed to meet the survey’s objectives by providing an estimate for a larger number of indicators at national/ area / sub-population level, for urban and rural areas together with four ecological zones: lowlands, foothills, mountains, and Senqu River Valley [11]. In stage one, urban, peri-urban areas within each district were identified as the main sampling strata (primary sampling units (PSUs); and in stage two, a sample of households was selected [11]. Within each stratum, a specific number of the Lesotho Census of Population and Housing of 2016 enumeration areas (EAs) were selected systematically with the probability proportional to size [11]. After the household listing was carried out within the selected EAs, a sample of 26 households was drawn in each sample EA (LMICS, 2018). However, in the data peri-urban strata were treated as rural to allow for comparability with previous surveys [11]. It was calculated that a minimum of 36 sample clusters were selected in each district [11]. There was an unequal allocation of the sample size within all ten districts- where some districts 42 clusters were collected in each district with a sample size of 10 400 households (400 clusters and 26 sample households per cluster) [11]. 10 413 households were sampled and only 8847 were interviewed (95.9% response rate). Among women aged 15-49, 7197 women were sampled and 6453 were interviewed (89.7% response rate), men (aged 15-49) 3417 were sampled but 2873 were interviewed with 84.1% response rate. Moreover, children under-five had a response rate of 91.2% and children aged 5-17 had a response rate of 94.0% [11]. For this study, the sample of children under-five is 3256.
Data variables
Dependent variables
The dependent variables are determined by the anthropometric Z-scores assigned to children in surveys. The assignment of anthropometric Z-Scores is based on the WHO Child Growth Standards that are developed through interpolation functions that take into consideration the sex, age, height (in centimeters) and weight of the child (in kilograms) [14]. Children are considered stunted when their height-for-age falls below negative two standard deviations of the median of the reference population [11, 14].
Independent Variable
This study follows the UNICEF framework for analyzing factors associated with child health and nutrition [11]. There are three levels predictors namely: immediate, underlying, and basic variables. Immediate variables are child sex, weight at birth and diarrhea as well as respiratory infection. At household level, underlying variables are households’ size, place of residence, female-headed households, household wealth, maternal age, maternal education as well as water sources and sanitation facilities. Moreover, at a community level, basic variables are community health seeking behavior, child immunization and antenatal care. Community food security, poverty, the proportion of mothers and males in the households with at least secondary, as well as communities with high proportions of female-headed households, and community media exposure, safe drinking water, and inadequate toilet facilities.
The variables were constructed as follows, for child level factors, child sex is binary (male and female), child weight at birth is categorized into three categories (Low birth weight (Less than 2.5kg), average weight (between 2.6kg and 3.8kg) and above average weight (greater than 3.8kg). For diarrhea and respiratory infections two weeks before the survey was categorized into three categories (yes, no and I don’t know). Among household variables, household size is categorized into 2-5 and 5+ place of residence (rural and urban), maternal residential status (yes and no), and female-headed households is also binary (yes and no). Household wealth was categorized into poorest, second, middle, fourth and richest wealth quintiles, maternal age (15-24, 25-34, 35+), maternal education (primary/and or no education, secondary and highest education). Moreover, safe water sources (safe and unsafe water) and adequate toilet facilities (adequate and inadequate toilet facilities. In relation to basic community variables, all variables are categorized as low and high proportions, low proportions are clusters with proportions less than 40% and high proportions are clusters with proportions more than 40%.
Data Analyses
Data analysis is conducted in three steps, descriptive analyses, bivariate and multilevel logistic regression. Descriptive analysis of all the independent variables involves, frequencies, percentages, confidence intervals to determine statistical differences and p-values. Secondly, a chi-square analysis (bivariate analysis) is conducted to determine which variables are statistically significant to be included into the main model at 95% confidence internal (p-values <0.05). A multilevel logistic regression analysis is carried out because the data has evidence of clustering and hierarchy. In multi-level research, the structure of data in the population is hierarchical, and a sample for such a population can be viewed as a multistage sample [15]. With this explanation, the data analysis model best suits the MICS dataset because it was nested in two stages. The units at lower-level (level-1) are individual and clusters are again nested within units at the next higher level- which is level 2. This kind of clustering can introduce multi-level dependency or correlation among the observations that can have implication for model parameter estimates [15].
To measure the dependency in the data a three-stage multilevel analysis is conducted by running an empty model and calculating the intra-class correlation coefficient (ICC). The aim of the empty model is to find log-odds of the dependent variable while including no predictors [16]. Secondly, by running a model that will measure the effects of lower-level variables because these intermediate variables are allowed to vary from one cluster to another. Thirdly, running a model with level-2 predictors and level-3 predictors. Finally, by running the final model that includes individual, household, and community level variables. The statistical analyses are carried out in this study with the help of STATA15 software.
Profile of the Study Population
Table 1 shows all that majority of children in the sample did not receive minimal acceptable diet (MAD) the day before the survey (68.0%), aged 6-8 months (73.7%), and there was no significant difference between boys and girls (49% vs 50%). Moreover, majority of them were born with birth weight greater than 3.9kg (66.44%) and did not have diarrhea (91.4%) and respiratory infection (59.4% two weeks before the survey. At household level, majority of them were from households sized 5+ (61.0%), with residential mothers (85.3%), female headed households (59.8%), poorest households (22.7%), maternal education of no education and/or primary education (53.1%) as well as maternal age of 25-34 (47.1%). They were also from households with access to safe drinking water (82.3%) and adequate toilet facilities (70.0%). In relation to communities, majority of them were from communities with low proportions of health seeking behavior (68.9%), immunization (67.4%), males in households with at least secondary education (54.4%), female headed households (85.5%), poor households (61.0%) and from communities with low proportions of maternal media exposure (78.8%). Additionally, majority of them were from communities with high proportion of antenatal care (57.3%), food secure households (households that owned livestock and land) (70.0%), household with maternal education with at least secondary education (68.4%) and households with safe drinking water sources (81.5%) as well adequate toilet facilities (64.4%).
Table 1: Characteristics of the study population: Lesotho 2018 (N= 3256)
Immediate Variables
|
N
|
Percentage
|
Dietary Intake
|
|
|
No MAD
|
814
|
68.0
|
MAD
|
366
|
32.0
|
Child Age
|
|
|
6-8
|
2399
|
73.7
|
9-23
|
857
|
26.3
|
Child Sex
|
|
|
Male
|
1595
|
49.0
|
Female
|
1661
|
51.0
|
Child Weight at Birth
|
|
|
Less than 2.8kg
|
166
|
5.1
|
2.6kg to 3.8kg
|
928
|
28.5
|
≥3.9kg
|
2163
|
66.4
|
Diarrhea
|
|
|
Yes
|
281
|
8.6
|
No
|
2975
|
91.4
|
Respiratory Infection
|
|
|
Yes
|
1324
|
40.6
|
No
|
1934
|
59.4
|
Underlying Variables
|
Household Size
|
|
|
2-5
|
1271
|
39.0
|
5+
|
1985
|
61.0
|
Place of Residence
|
|
|
Urban
|
1328
|
40.8
|
Rural
|
1928
|
59.2
|
Maternal Residential Status
|
|
|
Yes
|
2778
|
85.3
|
No
|
478
|
14.7
|
Female Headed Households
|
|
|
Yes
|
1946
|
59.8
|
No
|
1311
|
40.2
|
Household Wealth
|
|
|
Poorest
|
738
|
22.7
|
Second Wealth Quintile
|
702
|
21.6
|
Middle Wealth Quintile
|
662
|
20.3
|
Fourth Wealth Quintile
|
599
|
18.4
|
Richest
|
555
|
17.0
|
Maternal Age
|
|
|
15-24
|
833
|
33.2
|
25-34
|
1184
|
47.1
|
35+
|
494
|
19.7
|
Maternal Education
|
|
|
Primary or none
|
1381
|
53.1
|
Secondary
|
275
|
10.6
|
Beyond secondary
|
945
|
36.3
|
Sources of Drinking Water
|
|
|
Unsafe
|
575
|
17.7
|
Safe
|
2681
|
82.3
|
Toilet Facilities
|
|
|
Inadequate
|
978
|
30.0
|
Adequate
|
2278
|
70.0
|
Basic Variables
|
Community Health Seeking Behavior
|
|
|
Low
|
2244
|
68.9
|
High
|
1012
|
31.1
|
Community Immunization
|
|
|
Low
|
2195
|
67.4
|
High
|
1061
|
32.6
|
Community Antenatal Care
|
|
|
Low
|
128
|
12.7
|
High
|
876
|
87.3
|
Community Food Security
|
|
|
Low
|
976
|
30.0
|
High
|
2280
|
70.0
|
Community Maternal Education
|
|
|
Low
|
1030
|
31.6
|
High
|
2226
|
68.4
|
Community Male Education
|
|
|
Low Proportion
|
1772
|
54.4
|
High Proportion
|
1484
|
45.6
|
Community Female-Headed Households
|
|
|
Low
|
2783
|
85.5
|
High
|
473
|
14.5
|
Community Poverty
|
|
|
Low
|
1985
|
61.0
|
High
|
1271
|
39.0
|
Community Media Exposure
|
|
|
Low
|
2567
|
78.8
|
High
|
689
|
21.2
|
Community Safety of Sources of Drinking Water
|
|
|
Low
|
603
|
18.5
|
High
|
2653
|
81.5
|
Community Adequacy of Toilet Facilities
|
|
|
Low
|
1158
|
35.6
|
High
|
2098
|
64.4
|
Bivariate Analysis
Table 2 presents all three level factors and their association (chi square p-value) with stunting at bivariate analysis. In Lesotho, a third (33.6%) of under5s were stunted. All variables with a p-value less than 0.05 from Chi-Square were considered significantly associated with stunting. Dietary intake, child weight at birth, and respiratory infections were immediate variables significantly associated with stunting. Underlying variables associated with stunting were place of residence, households’ wealth index, maternal education and residential status, water sources and toilet facilities. For basic community variables: community immunication rates community maternal and male education, community food security, community drinking water sources safety, community toilet facilities adequacy and community media exposure were significantly associated with stunting. All variables that were statistically significant in Table 2 were further tested using an Adjusted Wald Statistics where one variable (community immunization) was excluded in the main model.
Table 2: Prevalence of Stunting among Children Under Five: Lesotho 2018 (N= 3256)
Factors
|
Not Stunted
|
Stunted
|
P-Value
|
%
|
N
|
CI
|
%
|
N
|
CI
|
Immediate variables
|
Dietary Intake
|
|
|
|
|
|
|
|
No MAD
|
59.1
|
402
|
(54.8,63.4)
|
40.9
|
278
|
(36.6,45.2)
|
0.011
|
MAD
|
70.4
|
141
|
(62.6,77.2)
|
29.6
|
59
|
(22.8,37.4)
|
Child Sex
|
|
|
|
|
|
|
|
Male
|
64.5
|
1028
|
(61.1,67.7)
|
35.5
|
567
|
(32.3,38.9)
|
0.085
|
Female
|
68.2
|
1132
|
(65.6,70.7)
|
31.8
|
529
|
(29.3,34.5)
|
Child Weight at Birth
|
|
|
|
|
|
|
|
< 2.5kg
|
47.4
|
79
|
(2.8, 4.8)
|
52.6
|
87
|
(27.1,32.1)
|
0.000
|
2.6 kg - 3.8kg
|
68.8
|
638
|
(6.1 10.9)
|
31.2
|
290
|
(23.6,29.5)
|
> 3.8kg
|
66.8
|
1444
|
(4.2,6.2)
|
33.2
|
719
|
(26.5,30.6)
|
Diarrhea
|
|
|
|
|
|
|
|
Yes
|
66.5
|
187
|
(59.4,73.0)
|
33.5
|
94
|
(27.0,40.6)
|
0.963
|
No
|
66.3
|
1966
|
(64.1,68.5)
|
33.7
|
996
|
(31.5,35.9)
|
Respiratory Infection
|
|
|
|
|
|
|
|
Yes
|
62.8
|
831
|
(59.6,66.0)
|
37.2
|
492
|
(34.0,40.5)
|
0.002
|
No
|
68.7
|
1320
|
(66.3,71.1)
|
31.3
|
601
|
(28.9,33.7)
|
Underlying variables
|
Household Size
|
|
|
|
|
|
|
|
2-5
|
67.0
|
852
|
(63.3,70.5)
|
33.0
|
420
|
(29.5,36.7)
|
0.644
|
5+
|
66.0
|
1309
|
(63.4,68.5)
|
34
|
676
|
(31.6,36.7)
|
Place of Residence
|
|
|
|
|
|
|
|
Urban
|
72.1
|
958
|
(68.6,75.4)
|
27.9
|
370
|
(24.6,31.4)
|
0.000
|
Rural
|
62.4
|
1201
|
(59.8,64.9)
|
37.6
|
726
|
(35.1,40.2)
|
Household Heads
|
|
|
|
|
|
|
|
Male
|
63.2
|
1314
|
(65.0,70.0)
|
36.8
|
632
|
(30.0 35.0)
|
0.156
|
Female
|
64.3
|
846
|
(61.2,67.9)
|
35.7
|
463
|
(32.1,38.8)
|
Household Wealth
|
|
|
|
|
|
|
|
Poorest
|
55.7
|
412
|
(52.2,59.2)
|
44.3
|
327
|
(40.8,47.8)
|
0.000
|
Second
|
62.3
|
438
|
(58.0,66.5)
|
37.7
|
265
|
(33.5,42.0)
|
Middle
|
65.6
|
434
|
(60.5,70.3)
|
34.4
|
228
|
(29.7,39.5)
|
Fourth
|
71.0
|
425
|
(65.1,76.2)
|
29.0
|
174
|
(23.8,34.9)
|
Richest
|
81.5
|
452
|
(76.3,85.8)
|
18.5
|
103
|
(14.2,23.7)
|
Maternal Age
|
|
|
|
|
|
|
|
15-24
|
64.0
|
534
|
(60.2,67.7)
|
36.0
|
300
|
(32.3,39.8)
|
0.266
|
25-34
|
68.3
|
808
|
(64.9,71.5)
|
31.7
|
375
|
(28.5,35.1)
|
35+
|
66.8
|
330
|
(64.3,68.8)
|
33.2
|
164
|
(28.0,38.8)
|
Mother’s Residential Status
|
|
|
|
|
|
|
Yes
|
67.3
|
1868
|
(64.9,69.6)
|
32.7
|
909
|
(30.4,35.1)
|
0.008
|
No
|
61.1
|
293
|
(57.0,65.0)
|
38.9
|
186
|
(35.0 43.0)
|
Maternal Education
|
|
|
|
|
|
|
|
Primary or None
|
61.2
|
578
|
(55.7,64.5)
|
38.8
|
367
|
(35.5,42.3)
|
0.000
|
Secondary
|
67.1
|
926
|
(63.7,70.3)
|
32.9
|
454
|
(29.7,36.3)
|
Beyond secondary
|
82.6
|
227
|
(75.0,88.3)
|
17.4
|
48
|
(11.7,25.1)
|
Safety of drinking water
|
|
|
|
|
Unsafe
|
60.4
|
347
|
(55.6,65.0)
|
39.6
|
228
|
(35.0,44.4)
|
0.007
|
Safe
|
67.6
|
1813
|
(65.3,70.0)
|
32.4
|
868
|
(30.1,34.7)
|
Toilet Facilities adequacy
|
|
|
|
|
|
|
Inadequate
|
59.3
|
580
|
(55.7,62.9)
|
40.7
|
398
|
(37.1,44.3)
|
0.000
|
Adequate
|
69.4
|
1580
|
(66.6,72.0)
|
30.6
|
698
|
(28.0,33.4)
|
Basic variables
|
Health Seeking Behavior (%)
|
|
|
|
|
|
|
Low
|
65.9
|
1478
|
(63.3,68.3)
|
34.1
|
766
|
(31.7,36.7)
|
0.558
|
High
|
67.4
|
682
|
(63.0,71.5)
|
32.6
|
330
|
(28.5,37.0)
|
Immunization (%)
|
|
|
|
|
|
|
Low
|
66.0
|
1449
|
(63.2,68.6)
|
34.0
|
747
|
(31.4,36.8)
|
0.005
|
High
|
67.1
|
712
|
(63.6,70.4)
|
32.9
|
349
|
(29.6,36.4)
|
Antenatal Care (%)
|
|
|
|
|
|
|
Low
|
60.2
|
78
|
(47.8,72.1)
|
39.4
|
50
|
(27.9,52.2)
|
0.600
|
High
|
64.0
|
561
|
(60.4,67.4)
|
36.0
|
315
|
(32.6,39.6)
|
Community Food Security (%)
|
|
|
|
|
|
|
Low
|
67.6
|
1497
|
(68.5,76.4)
|
32.4
|
718
|
(23.6,31.5)
|
0.054
|
High
|
64.1
|
663
|
(61.2,66.1)
|
35.9
|
372
|
(33.9,38.8)
|
Community Maternal Education (%)
|
|
|
|
|
|
Low
|
60.3
|
621
|
(57.2,63.4)
|
39.7
|
408
|
(36.6,42.8)
|
0.000
|
High
|
69.1
|
1539
|
(66.3,71.8)
|
30.9
|
687
|
(28.2,33.7)
|
Community Male Education (%)
|
|
|
|
|
|
Low
|
61.8
|
1096
|
(59.0,64.5)
|
38.2
|
677
|
(35.5,41.0)
|
0.000
|
High
|
71.8
|
1065
|
(68.4,74.9)
|
28.2
|
419
|
(25.1,31.6)
|
Female Headed Communities (%)
|
|
|
|
|
|
Low
|
66.1
|
1840
|
(64.0,68.2)
|
33.3
|
943
|
(31.8,36.0)
|
0.642
|
High
|
67.7
|
320
|
(61.0,73.8)
|
32.3
|
153
|
(26.2,39.0)
|
Community Poverty
|
|
|
|
|
|
|
|
Low
|
71.5
|
1419
|
(68.4,74.3)
|
28.5
|
567
|
(25.7,31.6)
|
0.000
|
High
|
58.4
|
742
|
(55.1,61.5)
|
41.6
|
529
|
(38.5,44.9)
|
Community Media Exposure
|
|
|
|
|
|
|
Low
|
62.9
|
1615
|
(60.5,65.3)
|
37.1
|
952
|
(34.7,39.5)
|
0.000
|
High
|
79.1
|
546
|
(74.0,83.4)
|
20.9
|
144
|
(16.6,26.0)
|
Community Sources of Drinking Water safety
|
|
|
|
|
Low
|
60.8
|
367
|
(55.6,65.8)
|
39.2
|
236
|
(34.2,44.4)
|
0.014
|
High
|
67.6
|
1794
|
(65.3,69.8)
|
32.2
|
859
|
(30.2,34.7)
|
Community Toilet Facilities adequacy
|
|
|
|
|
|
Low
|
60.2
|
697
|
(56.9,63.4)
|
39.8
|
461
|
(36.6,43.2)
|
0.000
|
High
|
69.8
|
1464
|
(66.8,72.6)
|
30.2
|
634
|
(27.4,33.3)
|
Multilevel Model
The empty model (null model) was run to determine clustering, the second model included immediate and underlying variables (level 1 and level 2) and the third model included basic variables (level 3). In this study, there was evidence of variability in clustering. The null model had a Chi-Square of 19.91 and a p-value of 0.0000 making it statistically significant thus indicating clustering in the data. The Interclass correlation (ICC) of this model was right at the cut-off point at 0.054. Heck et al., 2014 discussed that 0.05 is often regarded as a conventional threshold to indicate more substantial evidence of clustering [17]. Moreover, the probability of being stunted in each community was (odd of being stunted/ (1+ odds of being stunted) 0.302. In general, the unconditional probability of a child being stunted is 30.2%. There was also variability of clustering between households and communities with a chi-square of 36.33 and p-value of 0.000 and ICC of 0.2574 (above the threshold).
Factors associated with stunting
Table 3 presents the main model of level one (immediate variables), level two (underlying variables) and level three (basic variables). At individual level, the odds are lower for children that did not receive MAD (WAOR=0.52; CI: 0.3, 0.9), born with greater than 3.8kg birth weight (WAOR=0.51; CI: 0.4, 0.6), and those that did not have respiratory infections two weeks before the survey (WAOR=0.61; CI: 0.4, 1.0) compared to their counterparts. At household level, the likelihood of stunting was lowest for education beyond secondary (WAOR=0.26; CI: 0.2, 0.4), fifth household wealth (WAOR=0.34; CI: 02, 03), safe sources of drinking water (WAOR=0.72; CI: 06, 09) and inadequate toilet facilities (WAOR=0.62; CI: 0.5, 0.7) compared to their counterparts. Higher odds were observed among Children from rural areas (WAOR=1.95; CI: 1.3, 2.1), mothers not residing within the household (WAOR=1.30; CI: 1.1, 1.6) compared to their counterparts. At community level, decreased odds were associated with children from communities with high community maternal education (WAOR=0.69; CI: 0.6, 0.8) and community male education (WAOR=0.56; CI: 0.5, 0.7) as well as those in communities with low safety of sources of drinking water (WAOR=0.73; CI: 0.3, 0.5), adequate toilet facilities (WAOR=0.66; CI: 0.5, 0.8) and high maternal media exposure (WAOR=0.37; CI: 0.3, 0.5) compared counterparts. Children from communities with high community poverty were two times (WAOR=2.04; CI: 1.7, 2.5) more likely to be stunted.
Table 3: Immediate, underlying and community factors associated with stunting: Lesotho 2018
Immediate Variables
|
UAOR (95% CI)
|
WAOR (95% CI)
|
WA (P-value)
|
Dietary Intake
|
MAD intake
|
|
|
|
No MAD (RC)
|
1.00
|
1.00
|
1.00
|
MAD
|
0.45 (0.2, 0.9)
|
0.52 (0.3,0.9)
|
0.027
|
Child Weight at Birth
|
< 2.6kg (RC)
|
1.00
|
1.00
|
1.00
|
2.6kg - 3.8kg
|
0.34 (0.1,0.9)
|
0.76 (0.6,0.9)
|
0.000
|
≥3.9kg
|
0.35 (0.1,0.9)
|
0.51 (0.4,0.6)
|
0.000
|
Respiratory Infection
|
Yes (RC)
|
1.00
|
1.00
|
1.00
|
No
|
0.86 (0.5,1.5)
|
0.61 (0.4,1.0)
|
0.004
|
Underlying Variables
|
Place of Residence
|
Urban (RC)
|
1.00
|
1.00
|
1.00
|
Rural
|
0.75 (0.4,1.3)
|
1.95 (1.3,2.1)
|
0.000
|
Household Wealth
|
Poorest (RC)
|
1.00
|
1.00
|
1.00
|
Second
|
0.85 (0.5,1.5)
|
0.73 (0.6,0.9)
|
0.013
|
Middle
|
0.71 (0.4,1.4)
|
0.55 (0.4,0.7)
|
0.000
|
Fourth
|
0.36 (0.1,1.0)
|
0.38 (0.3,0.5)
|
0.000
|
Richest
|
0.22 (0.6,0.8)
|
0.24 (0.2,0.3)
|
0.000
|
Maternal Residential Status
|
Yes (RC)
|
1.00
|
1.00
|
1.00
|
No
|
0.86 (0.5,1.5)
|
1.30 (1.1,1.6)
|
0.034
|
Maternal Educational Attainment
|
Primary or None (RC)
|
1.00
|
1.00
|
1.00
|
Secondary
|
0.94 (0.6,1.5)
|
0.73 (0.6,0.9)
|
0.003
|
Beyond secondary
|
2.11 (0.7,6.1)
|
0.26 (0.2,0.4)
|
0.000
|
Safety of Drinking Water
|
Unsafe (RC)
|
1.00
|
1.00
|
1.00
|
Safe
|
0.95 (0.6,1.6)
|
0.72 (0.6,0.9)
|
0.013
|
Adequacy of Toilet Facilities
|
Inadequate (RC)
|
1.00
|
1.00
|
1.00
|
Adequate
|
0.97 (0.6,1.6)
|
0.62 (0.5,0.7)
|
0.000
|
Basic Variables
|
Community Female Education
|
Low (RC)
|
1.00
|
1.00
|
1.00
|
High
|
0.91 (0.7,0.1)
|
0.69 (0.6,0.8)
|
0.000
|
Community Male Education
|
Low (RC)
|
1.00
|
1.00
|
1.00
|
High
|
0.95 (0.7,1.3)
|
0.56 (0.5,0.7)
|
0.000
|
Community Poverty
|
Low (RC)
|
1.00
|
1.00
|
1.00
|
High
|
1.54 (1.2,2.0)
|
2.04 (1.7,2.5)
|
0.000
|
Community Safety of Drinking Water
|
Low (RC)
|
1.00
|
1.00
|
1.00
|
High
|
0.86 (0.7,1.1)
|
0.73 (0.6,0.9)
|
0.017
|
Community Toilet Facilities
|
Low
|
1.00
|
1.00
|
1.00
|
High
|
0.94 (0.7,1.2)
|
0.66 (0.5,0.8)
|
0.000
|
Female Community Media Exposure
|
Low (RC)
|
1.00
|
1.00
|
1.00
|
High
|
0.51 (0.4,0.7)
|
0.37 (0.3,0.5)
|
0.000
|
Notes: MAD denoted Minimum Acceptable Diet, UAOR Unadjusted OR, WAOR, Wald Adjusted OR and WA Wald Adjusted